SVM Based Sentiment Analysis for Online Shopping Reviews

Today, ease of availability of internet has shifted time-consuming activities such as shopping online because it is a lot easier, comfortable, affordable and competitive as compared to traditional shopping. The only hassle though is that before shopping online, it has become important to read available reviews about the products and vendors as they play a vital role in decision-making. In most e-commerce websites, there are thousands of reviews published by consumers about specific products which makes it very difficult to analyze them for specific details. Therefore, this research intends to solve this problem through use of SVM classifier and proposes a system aimed to help the customer in their decision-making. In this paper, a system named “WISA Reviews and Feedbacks Analyzer” has been proposed. In the system, reviews are crawled from eBay's website then the dataset is cleaned and reviews are classified into positive, negative, and neutral categories. Users are also able to attach previously procured dataset files and the system classifies them producing results in form of charts and diagrams. The system uses Linear SVM algorithm and depicted better accuracy when compared with other similar classifiers.